Yongsen Zheng, Ruilin Xu, Ziliang Chen, Guohua Wang, Mingjie Qian, Jinghui Qin, Liang Lin. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Conversational Recommender System (CRS)aims to explicitly acquire user preferences towards items and attributes through natural language conversations.However, existing CRS methods ask users to provide explicit answers (yes/no) for each attribute they require, regardless of users' knowledge or interest, which may significantly reduce the user experience and semantic consistency.Furthermore, these methods assume that users like all attributes of the target item and dislike those unrelated to it, which can introduce bias in attributelevel feedback and impede the system's ability to accurately identify the target item.To address these issues, we propose a more realistic, user-friendly, and explainable CRS framework called Hierarchical User-Interest Tracking for Conversational Recommender System (HutCRS).HutCRS portrays the conversation as a hierarchical interest tree that consists of two stages.In stage I, the system identifies the aspects that the user prefers while the system asks about attributes related to these positive aspects or recommends items in stage II.In addition, we develop a Hierarchical-Interest Policy Learning (HIPL) module to integrate the decision-making process of which aspects to ask and when to ask about attributes or recommend items.Moreover, we classify the attribute-level feedback results to further enhance the system's ability to capture special information, such as attribute instances that are accepted by users but not presented in their historical interactive data.Extensive experiments on four benchmark datasets demonstrate the superiority of our method.The implementation of HutCRS is publicly available at https://github.com/xinle1129/HutCRS.

Heart failure (HF) has a highly variable annual mortality rate and there is an urgent need of determining patient prognosis to enable informed decision-making about heart failure treatment strategies. Existing survival risk prediction models either require features that limit their applicability or pose difficulties for parameter estimation as physicians have to use a limited set of variables with known hazard ratios published in literature. We propose a new model to predict the probabilistic survival score after HF diagnosis based on all clinical variables derived from the electronic health record (EHR). We formalize the parameter estimation problem by using the maximum likelihood estimation (MLE) principle and devise an effective and efficient algorithm to solve the optimization problem. Experimental results using EHR data of 234 HF patients validate the superiority of this new model in predicting prognosis over the currently used Seattle Heart Failure Model.

Feature-based collaborative filtering models, such as state-of-the-art factorization machines and regression-based latent factor models, rarely consider features' structural information, ignoring the heterogeneity of inter-type and intra-type relationships. Naïvely treating all feature pairs equally would potentially deteriorate the overall recommendation performance. In addition, human prior knowledge and other hierarchical or graphical structures are often available for some features, e.g., the country-state-city hierarchy for geographic features and the topical taxonomy for article features. It is a challenge to utilize the prior knowledge to further boost performance of state-of-the-art models. In this paper we employ rich features from both user and item sides to enhance latent factors learnt from interaction data, uncovering hidden structures from features' relationships and learning sparse pairwise and tree structural connections among features. Our framework borrows the modeling strengh from both structural sparsity modeling and latent factor models. Experiments on a real-world large-scale recommendation data set demonstrated that the proposed model outperforms several strong state-of-the-art baselines.

Unsupervised feature selection is a useful tool for reducing the complexity and improving the generalization performance of data mining tasks. In this paper, we propose an Adaptive Unsupervised Feature Selection (AUFS) algorithm with explicit l2/l0-norm minimization. We use a joint adaptive loss for data fitting and a l2/l0minimization for feature selection. We solve the optimization problem with an efficient iterative algorithm and prove that all the expected properties of unsupervised feature selection can be preserved. We also show that the computational complexity and memory use is only linear to the number of instances and square to the number of clusters. Experiments show that our algorithm outperforms the state-of-the-arts on seven different benchmark data sets.

Unlabeled high-dimensional text-image web news data are produced every day, presenting new challenges to unsupervised feature selection on multi-view data. State-of-the-art multi-view unsupervised feature selection methods learn pseudo class labels by spectral analysis, which is sensitive to the choice of similarity metric for each view. For text-image data, the raw text itself contains more discriminative information than similarity graph which loses information during construction, and thus the text feature can be directly used for label learning, avoiding information loss as in spectral analysis. We propose a new multi-view unsupervised feature selection method in which image local learning regularized orthogonal nonnegative matrix factorization is used to learn pseudo labels and simultaneously robust joint $l_{2,1}$-norm minimization is performed to select discriminative features. Cross-view consensus on pseudo labels can be obtained as much as possible. We systematically evaluate the proposed method in multi-view text-image web news datasets. Our extensive experiments on web news datasets crawled from two major US media channels: CNN and FOXNews demonstrate the efficacy of the new method over state-of-the-art multi-view and single-view unsupervised feature selection methods.

Problems of ordinal regression arise in many fields such as information retrieval, data mining and knowledge management. In this paper, we consider ordinal regression in a semi-supervised scenario, i.e., we try to utilize the ordinal information from the distribution of unlabeled data. Semi-supervised ordinal regression is more applicable than traditional supervised ordinal regression, because nowadays labeled data is expensive and time-consuming as it needs human labor, whereas a large amount of unlabeled data are far accessible with the development of internet technology. We construct a general semi-supervised ordinal regression framework to formulate this problem. Based on the framework, we then propose a semi-supervised ordinal regression method called Semi-supervised Ordinal SVM (SOSVM). Additionally, in order to make our proposed method more applicable to problems with large scaled labeled data, we put forward a kernel based dual coordinate descent algorithm to efficiently solve SOSVM. Both rigorous theoretical analysis and promising experimental evaluations on real world datasets show the great performance and remarkable efficiency of SOSVM.

Aggregate outputs learning differs from the classical supervised learning setting in that, training samples are packed into bags with only the aggregate outputs (labels for classification or real values for regression) known. This setting of the problem is associated with several kinds of application background. We focus on the aggregate outputs classification problem in this paper, and set up a manifold regularization framework to deal with it. The framework can be of both instance level and bag level for different testing goals. We propose four concrete algorithms based on our framework, each of which can cope with both binary and multi-class scenarios. The experimental results on several datasets suggest that our algorithms outperform the stateof-art technique.

Semi-supervised learning has been successfully applied to many fields such as knowledge management, information retrieval and data mining as it can utilize both labeled and unlabeled data. In this paper, we propose a general semi-supervised framework for multi-class categorization. Many classical supervised and semi-supervised method dealing with binary classification or multi-class classification including the standard regularization and the manifold regularization can be viewed as special cases of this framework. Based on this framework, we propose a novel method called multi-class unlabeled constrained SVM(MCUCSVM) and its special case: multi-class Laplacian SVM(MCLapSVM). We then put forward a general kernel version semi-supervised dual coordinate descent algorithm to efficiently solve MCUCSVM and makes it more applicable to problems with large number of classes and large scale labeled data. Both rigorous theory and promising experimental results on four real datasets show the great performance and remarkable efficiency of MCUCSVM and MCLapSVM.

Inspired by the fact that the final decision rule is mainly affected by a small subset of the training samples, i.e., Support Vector Machine (SVM) shows that the decision function relies on the few samples that are on or over the margin. We propose a new framework that explicitly strengthen this intuitive fact by adding an l1-norm regularizer. We give different formulations for our framework in different scenarios, and the experiments show that our framework can not only lead to high sparse solutions but also better performance than traditional methods.

Based on the theory of electrostatic field, a novel semi-supervised learning method named Poisson propagation has been proposed by Fei Wang. In his formulation, data are regarded as points in the field and the labels of unlabeled data points are propagated from labeled sources, which is like the field responses modeled by Poisson's equation. In this paper, we develop an efficient way for accelerating the PP algorithm, and also provide the theoretical analysis of the optimality of such acceleration approach. Our method is tested on 6 different data sets. The experiment results show the effectiveness of our acceleration algorithm.